Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
GC51D-1024 GC51D-1024 2011 AGU Fall Meeting 2011 AGU Fall Meeting HIGH AND LOW RAINFALL EVENTS IN HAWAI‘I IN RELATION TO LARGE-SCALE CLIMATE ANOMALIES IN THE PACIFIC Mami Takahashi1, Oliver Elison Timm2, Thomas W. Giambelluca1, and Henry F. Diaz3 1Department of Geography, University of Hawai‘i at Mānoa, HI, USA 2International Pacific Research Center, University of Hawai‘i at Mānoa, HI, USA 3NOAA/ESRL/CIRES, University of Colorado, Boulder, CO, USA Email: Takahashi: [email protected]; Elison Timm: [email protected] Giambelluca: [email protected]; Diaz: [email protected] Heavy Rain Events Data & Methods We selected 12 stations with daily reported precipitation amounts between 1958–2005 (see map). This period was found to be most suitable based on data availability and homogeneity. The selected stations have the lowest numbers of missing observations (80–99% complete). We concentrate on the wet season months (October-April). Introduction Future climate change is expected to increase the frequency and intensity of extreme weather events in the coming decades. But the confidence is considerably low in projecting hydrological changes on a local scale directly from global climate models (GCMs). Here we present results form our regional statistical downscaling analysis of daily heavy rainfall events and low rainfall months in the Hawaiian Islands. The goal of our research: • To understand how large-scale circulation anomalies have affected the rainfall characteristics across the Hawaiian Islands during the 20th century. • Apply this information in statistical downscaling to project model scenario simulations from the IPCC reports. The major islands of Hawaii are located in the NE trade wind zone. For most regions, the bulk of the annual rainfall occurs during the months of November through April. Wettest areas are usually found along the windward slopes of mountains (maximum 11,000 mm/a). Dry regions are characterized by sporadic rainfall, with a few heavy rain events during the year contributing more than 50% to the annual totals. • Develop products that provide quantitative estimates for changes in the occurrence of heavy rain days and changes in the probability of very low rainfall months. Fig. 1 Trend from mid to late 20th century Kaneha, Kauai Composite anomalies for dry months at two stations: 700 hPa specific humidity (g/kg in colors) and 500 hPa geopotential (contours in m). We work with geopotential height in 500 hPa, winds in 1000 and 700 hPa, humidity and moisture transport (700 hPa) and temperature difference (1000 hPa minus 500 hPa). For each station a set of anomaly maps is obtained. The anomaly patterns are used to define a projection index that measures for each month the similarity between the composite pattern and the actual circulation. This method is applied to NCEP reanalysis data and model scenarios likewise. Fig. 2 Each station has its own projection index time series that is translated into a probability measuring the likelihood for a dry month. We calculated the ratio of the number of dry months versus the total months within different projection index ranges (bins) and fitted a piecewise linear function through the data points. 1st PCA mode: . Negative loading Positive loading Figure 1:Change in probability for dry months (precipitation < 10% quantile) 1977-2008 minus 1949-1976. 21st century future scenarios: Fig. 3 Heavy rain days SOI We selected 131 stations with monthly precipitation amounts between 1948–2007, using only wet season months November through April. Months with rainfall below the 10% quantile were identified and composite anomaly maps were derived from the NCEP reanalysis data. Lahaina, Maui . Figure (a): Number of heavy rain days per season for 1958-1976 (left circles) and 1977-2005 (right circles); (b) Regression based estimates using SOI and PNAI. PNAI Hawaii Results: Dry months analysis regression model SOI +PNAI observed Data & Methods Mean Annual Rainfall Results: Heavy rain analysis NCEP reanalysis 500 hPa and 1000 hPa geopotential height data (October-April season, 1958 - 2005) are used as large-scale circulation anomalies in this study. We regressed the 1000 hPa 500 hPa geopotential height anomalies onto the instrumental Southern Oscillation Index (SOI) and Pacific North American index (PNAI), respectively. These patterns (see below) are used to project the climate change scenario simulations onto these two dominant modes of North Pacific climate variability. Low Rainfall Months Rainfall in Hawaii SOI & PNAI Summary of 6 model simulations 2046-2065 (red) and 2081-2100 (pink) A1B and A2 together. Left: Smoothed histogram using estimated changes from all stations and all simulations. Middle: 20-yr mean average SOI and PNAI projections. Contours show the estimated number of heavy rain events for the 12-station average. Right: The mid-1970s climate shift is of similar magnitude as the projected future changes in SOI and PNAI. less more dry months Station sample neg. PCA loading pos. PCA loading Lahaina, Maui Figure 2: Principal component analysis (PCA) loading pattern for the low rainfall probability indices (131 stations). PCA was derived from the MPI ECHAM5 simulation 1980-2000. Fig. 4 Geospatial interpolation (experimental stage) Kaneha, Kauai Piecewise linear transfer function between multivariate circulation projection index and probability for dry months conditions. Acknowledgements The work reported here was supported by the Pacific Island Ecosystems Research Center (Biological Resources Discipline, USGS), and the US Fish and Wildlife Service through the Pacific Island Climate Change Cooperative (CA #12200-94023), and the U.S. Army Corps of Engineers Cooperative Agreement #W912HZ-11-2-0035. Oliver Elison Timm acknowledges the support by the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) through its sponsorship of the International Pacific Research Center. References: R Figure 3: Change in winter drought probability. Shown is the ratio R between the probability 2046-2065 and 1981-2000 (dashed) and 2081-2100 and 1981-2000 (solid line). Red lines show the smoothed histogram of R from samples using all stations with positive PCA loading (see Figure 2) and all scenarios. Blue lines show the density estimates from stations with negative PCA loading. Figure 4 shows a first map produced with spatial interpolation methods to project future drought risk areas for the Hawaiian Islands. • Elison Timm, O., H. F. Diaz, T. W. Giambelluca, and M. Takahashi, J. Geophys. Res.116, D04109, doi:10.1029/2010JD01492, 2011. • Giambelluca TW, Chen Q, Frazier AG, Price JP, Chen Y-L, Chu P-S, Eischeid J., and Delparte, D. 2011. The Rainfall Atlas of Hawai‘i. http://rainfall.geography.hawaii.edu. • Norton, C. W., P.-S. Chu, and T. A. Schroeder (2011 J. Geophys. Res., 116, D17110, doi:10.1029/2011JD015641.